We hope you'll join us for our 4/23 webinar on using data tables to apply reference ranges and AE codes in OC4. For more information and to register, visit https://register.gotowebinar.com/register/2882170018956684555

How do you track DM review in this database?

2

Comments

  • kristiakkristiak Posts: 1,329 ✭✭✭
    When we started the Scandinavian Simvastatin Survival  study in 1987 (4444 patients) we designed the first global ECD system with Digital Equipment.Inc. In that database we had the possibility to mark an individual. Already at this point in time we used the computers to help us with the review. We soon discovered that we did not need the status markers since the automatic queries did the job for us. Today we use the SDV status and run a slightly different set of queries and finally at database lock .we run the final query to pick up any outstanding items. I created the very first TDR system already in 1967 and it has been improved ever since. Today we use the excellent export functions created by Gerben Rienk Visser at Trial Data Solutions. His method creates a very long and skinny table were each data item with its meta data is one raw. From this table we create all the queries;, anything between 50 and 250 depending on the number of variables in the protocol 
    Believe me.... when you have billions of datapoints in a study, this is the only way to go! :)
    .  
  • kristiakkristiak Posts: 1,329 ✭✭✭
    I can understand if you have worked with RAVE for a long time, OC might seem a bit poor. But considering what you get for nothing for nothing in an open structure where you have all the experts in the user group around the world to help you create your own system that can be tailored to your own needs, nothing beets that!
  • lindsay.stevenslindsay.stevens Posts: 404 ✭✭✭
    via Email
    I agree with Krister, it is much more efficient and reliable to conduct
    data cleaning with programmed checks, either as rules or some code run
    against extracted data.

    A monitor will mark SDV or raise a query for example as a result of
    comparing recorded data to source.

    A data manager might want to do the same during review but what is the
    action they are performing? Really it is reviewing the data against a set
    of rules, which should be articulated into code.

    Then, the only action that a data manager needs to record is queries where
    the rule has failed, although this can be automated as well.

    So finally if all this is set up then all the data manager needs to do is
    ensure queries are resolved. The discrepancy notes interface does this
    nicely.

    As far as I'm aware the Enterprise licence doesn't have extra tools for
    this sort of thing, but there are plenty of community tools available, most
    of which are listed on the Extensions page.
  • kristiakkristiak Posts: 1,329 ✭✭✭
    Thanks for your support, Lindsay. I have not tested the data mart in the Enterprise version but I understand that it may make a manual review a little simpler by the ability to create queries in this inter phase.


  • juan.debonisjuan.debonis Posts: 36
    via Email
    Dear Kristiak,

    I bet for Clinical Data Managers, yours, is great story to hear and
    learn.

    We use a similar approach with a combination a available tools,
    internal development and strong support form standards (clinical, like
    CDISC, and technology, like XML): ODM + xQuery + csv (trying to avoid
    proprietary technologies when possible).

    XQuery has demonstrated us to be a really powerful language in
    order to develop complex reviews (we are surprised every day for the
    simplicity of "reading the code" to understand what it does). We also
    have a standard set of queries (based on CDISC standardizations when
    designing forms) and adding study customized set of queries.

    Juan.

    Juan De Bonis
    *3Eff S.A.*


    El 04/06/2015 a las 13:44, kristiak escibió:
  • kristiakkristiak Posts: 1,329 ✭✭✭
    Hi Juan,

    Seems that you have a great thing going there. Yes we follow the same approach and use. the CDISC ODM as well.  Keeps the auditors happy.Our queries we mainly write in SQL (I'm an old Oracle freak)  
    Best Regards
    Krister
  • ccollinsccollins Posts: 379 admin
    edited June 2015
    Hi,

    There is a Jira ticket requesting DM review as a feature: https://jira.openclinica.com/browse/OC-5211

    I've heard it requested before from users but it never has been a particularly high priority from anybody I've spoken with. It sounds like it's very important to a few of you, if so we'll raise its priority in the core system development queue. Perhaps others will want to help develop this feature too.

    It also sounds like there are more automated, elegant models for accomplishing the end goal, that could be done either within or external to OpenClinica. As Lindsay mentioned, there are a growing number of tools on the extensions site that may be useful (some of the best ones are the ones Lindsay has written :-) But I imagine having the capability built intrinsically into OpenClinica is ideal since it is, after all, a data management platform. 

    I think that leaves two questions:

    - @eemshwil and others, is the need for manual DM review 'flags' still important in light of the later comments? If so we will move this up the priority list in the backlog

    - What would a module in OC that automatically 'reviews data against a set of rules articulated into code' look like? Is the current rule engine a good starting point, if so what does it lack? Or does it require a different approach? What are the requirements and what UI/reports/management screens are needed? 

    Best,
    Cal







  • eemshwileemshwil Posts: 10
    edited June 2015
    Oh goodness, yes I completely agree with you all and this what I'm accustomed to - programming a database so that THE SYSTEM reviews the data for DM and we're not doing it all manually. However, I'm afraid certain situations do not always allow for this.  Instead of boring you with the detailed history, I'll just say that I've been given a legacy study that was very poorly built, very few edit checks written and so very little cleaning of data has been performed.  It would be very nice in this case to have a feature that helped me to keep track of my DM review.  I'm not sure why there are some saying this is an outdated process when mega software systems like Medidata RAVE use this feature?  Time wasted on guessing or tracking through the audit trail (or even adding some outside program to help track) if something has been DM reviewed seems silly to me - just put the feature directly in the software as I agree with ccollins - OC "is, after all, a data management platform."  Finding all these work arounds seems counterproductive. Also, there is no EDC system I know of that is complex enough to check the data logic between AEs, ConMeds, MedHx.  There must be SOME manual review done by DM and so with this - HOW is this manual review tracked by a DM department when they use OC that is not counterproductive/making more work for themselves?
  • eemshwileemshwil Posts: 10
    Cal - to answer your questions, my response is in blue below:

     @eemshwil and others, is the need for manual DM review 'flags' still important in light of the later comments? If so we will move this up the priority list in the backlog. Yes, very important and critical for several of our studies at this point in time.  

    - What would a module in OC that automatically 'reviews data against a set of rules articulated into code' look like? Is the current rule engine a good starting point, if so what does it lack? Or does it require a different approach? What are the requirements and what UI/reports/management screens are needed? A management screen much like the SDV Task in OC for DM review, would be great.  It would be important too that if data was edited at any point after DM review has been performed/flagged as such, that the flag be removed as this lends suggestion to the DM that data has been changed and will need to be re-reviewed - and if a query was to be set on that re-reviewed data point that the SDV flag be removed as well to ensure it's SDV'd yet again by the CRA.
  • ccollinsccollins Posts: 379 admin
    Thanks! I've added your comments to the ticket and moved it up on the priority queue.
This discussion has been closed.